A differential cargo loading model of ciliary length regulation by IFT

Summary

Background

During the assembly and maintenance of cilia, precursor proteins need to be transported from the cell body into the organelle. Intraflagellar transport (IFT) is assumed to be the predominant protein transport pathway in cilia but it remains largely unknown how ciliary proteins use IFT to reach their destination sites in the cilium and whether the amount of cargo transported by IFT is regulated.

Results

Single particle imaging showed that DRC4, a structural protein of the axoneme, moves in association with IFT particles inside Chlamydomonas reinhardtii cilia. IFT is required for DRC4 transport both into and within the cilium. DRC4 cargoes dissociate from IFT trains at the tip as well as at various sites along the length of the cilium. Unloaded DRC4 diffuses before docking at its axonemal assembly site. In growing cilia, DRC4 transport by IFT was strongly increased over the steady-state level and the frequency decreased linearly with the increasing ciliary length. The frequency of DRC4 transport was similarly elevated in short growth-arrested cilia and remained high even when the amount of DRC4 available in the cell body was reduced.

Conclusions

DRC4 is a bona fide cargo of IFT. Incompletely assembled cilia trigger an increase in the amount of DRC4 cargo transported by IFT particles and DRC4 transport is down-regulated as cilia approach their steady-state length. We propose a model in which ciliary length is controlled by regulating the amount of cargo transported by IFT.

Introduction

Cilia and flagella are microtubule-based extensions of cells that function in cell locomotion, fluid propulsion, and signaling. Defective cilia have been implicated in a wide range of human diseases including male infertility, blindness, polycystic kidney disease, and obesity [1]. Because cilia are devoid of protein synthesis, hundreds of different proteins need to be transported from the cell body into and within the organelle during ciliary assembly [2]. The ciliary axoneme grows by the addition of subunits to the distal tip suggesting that protein supply requires an active transport [3]. The assembly of most cilia is dependent on the conserved intraflagellar transport (IFT) pathway, the bidirectional movement of protein particles along the ciliary microtubules of the axonemes [4]. Anterograde IFT particles move from the cell body to the ciliary tip powered by kinesin-2, and after remodeling at the ciliary tip, the retrograde IFT particles are carried back to the cell body by dynein 2 [5–8]. IFT particles are composed of about two dozen distinct polypeptides and have been assumed to function as carriers for structural proteins of cilia. IFT is also required for the maintenance of mature cilia and for cilia-based signaling [9].

Twenty years after the first description of IFT [4], our knowledge of how IFT actually transports cargo proteins is still very limited. Several lines of evidence have indirectly suggested a role for IFT in the transport of structural proteins of the ciliary axoneme [10–14]. Recently, it has been shown that the IFT particle proteins IFT74 and IFT81 form a tubulin binding module [15]. Only a few cargo proteins, i.e. proteins that move with IFT velocity inside cilia but are not required for IFT itself, have been identified [16–19]. It remains unclear which proteins actually travel by IFT, how they move to their docking sites inside cilia, and how the amount of protein transported by IFT is regulated. Because IFT particles are likely to carry single copies or small clusters of a given cargo, a highly sensitive imaging technique is required to image individual proteins during transport. Here, we took advantage of the intrinsic behavior of C. reinhardtii cells to adhere to cover glass surfaces by the means of their two cilia (Fig. 1; we refer to the two flagella of C. reinhardtii as cilia because cilia and flagella are essentially identical organelles). This brings cilia (~200 nm diameter) into the range of the evanescence field generated by the reflected laser beam in Total Internal Reflection Fluorescence (TIRF) microscopy allowing to image single GFP-tagged proteins inside cilia [20, 21]. We analyzed ciliary transport and assembly of DRC4, a conserved bona fide axonemal protein of the nexin-dynein regulatory complex (N-DRC) which is located on the inside of the outer doublets [22–24]. The N-DRC linkers are critical for ciliary motility but dispensable for IFT and ciliary assembly. In C. reinhardtii, DRC4 is encoded by the PF2 gene; pf2 mutants have aberrant motility and defects in assembly of the N-DRC and associated inner arm dyneins [24–26]. Expression of DRC4-GFP using its own promoter restores wild-type ultrastructure and motility of cilia of pf2 null mutants, demonstrating that the tagged protein is functional and transported in near wild-type amounts into cilia [Fig. S1; [22, 23]]. Here, we show that DRC4-GFP is a cargo of IFT and that the frequency of DRC4-GFP transport is inversely proportional to the length of cilia.

Fig. 1. DRC4-GFP is a cargo of IFT.

Two cells expressing DRC4-GFP in DIC (a) and TIRF (b, c) before (b) and after photobleaching (c). Arrowheads: cilia. Bar = 2 µm. d–f, Kymograms showing anterograde IFT (arrows in d, e), diffusion (arrowheads in e), and retrograde IFT (arrow in f) of DRC4-GFP. Time progresses from left to right. Anterograde particles (base to tip) move from the lower left to upper right, retrograde particles (tip to base) move from upper left to lower right. g–k, Simultaneous imaging of DRC4-GFP (g, h) and IFT20-mCherry (i, j). Open arrowheads in g, cilia; filled arrowhead, position of the ciliary bases. h, j, k, kymograms for DRC4-GFP (h), IFT20-mCherry (j), and the merged colors (k). The IFT particles carrying DRC4-GFP are marked by arrows. Bars (d–k) = 1 s/2 µm.

Results

IFT traffics axonemal proteins

Cells expressing DRC4-GFP are shown in Fig. 1a, b. Cilia of DRC4-GFP cells were photobleached to remove the signal of DRC4-GFP already assembled into the axoneme; this allowed us to image DRC4-GFP transported inside cilia (Fig. 1 a–c). TIRF microscopy of living cells revealed DRC4-GFP that moves inside cilia by two distinct modes: diffusion and active transport (Figs. 1d–f; movies S1–S3). Kymograms, plots of particle position vs. time, are used to present the dynamic properties of IFT in a single image. In such plots, particles undergoing directional active transport will form a diagonal line with a slope indicating the particle velocity (Fig. 1d–f) whereas particles undergoing diffusion display in random back-and-forth trajectories (Fig. 1e). The anterograde transport of DRC4-GFP progressed with a velocity of ~1.9 µm/sec (STD 0.23 µm/sec; n=78; Fig. 1d). In steady-state cilia, the frequency of anterograde transport of DRC4-GFP was 0.65 particles/min (STD 0.5, n = 5 experiments with 10 to 65 measurements each). The frequency of anterograde IFT is ~60 particles/min indicating that in steady-state cilia about 1% of the IFT particles carry DRC4-GFP as a cargo. Retrograde transport of DRC4-GFP was less frequent (< 0.1 particles/min in steady-state cilia) with a velocity of ~3 µm/sec (STD 0.78 µm/sec; n=16; Fig. 1f). These velocities are characteristic for IFT in C. reinhardtii [4]. To test directly whether DRC4-GFP is transported by IFT, we analyzed a strain expressing both DRC4-GFP and an IFT particle protein, IFT20-mCherry, in a corresponding double mutant background (Fig. S1). Simultaneous imaging of both proteins revealed that DRC4-GFP co-localized with IFT20-mCherry during anterograde transport (n = 47 co-transportation events from 27 recordings; Fig. 1g–k; movies S4, S9). Thus, DRC4-GFP is a cargo of IFT. Similarly, we observed that DRC2-GFP, another component of the N-DRC, and the ciliary central pair protein PF16-GFP [22] are also transported by IFT (Fig. S2). Although long predicted, our data demonstrate directly that axonemal proteins travel within cilia as cargoes on IFT particles.

DRC4-GFP requires IFT for entry into cilia

Protein transport between the cell body and the cilium is believed to be controlled by the most proximal part of the cilium called the transition zone [27]. To image DCR4-GFP at the ciliary base, we used Chlamydomonas zygotes. Zygotes initially lack a cell wall; this reduces the distance between the cell apex and the cover glass. In favorable positioned cells a dot-like DRC4-GFP signal, presumably representing DRC4-GFP at the basal body, was observed at the base of each cilium (Fig.2 a, b). These signals were separated from the cilia by a narrow gap, which is likely to correspond to the transition zone. After photobleaching, DRC4-GFP was observed moving with linear trajectories indicative for active transport from the area of the basal bodies into the cilium proper (Fig. 2 c, d). The observation suggests that DRC4-GFP is carried on IFT particles from the cell body into the cilium. Soluble non-ciliary proteins up to a size of ~40 kD can enter cilia by diffusion [28], and Lin et al. [29] showed proteins of up to ~650 kD to diffuse into cilia. [28, 29]. Phospholipase D (~25 kD;[30]) and SAG1 (~65 kD;[31]), two proteins of the Chlamydomonas ciliary membrane, enter and accumulate in cilia in the absence of IFT, and the ciliary transmembrane proteins somatostatin receptor 3 and smoothened move predominantly by diffusion inside primary cilia [32]. To determine whether DRC4-GFP (~78 kD) can also enter cilia by diffusion, we expressed DRC4-GFP in the temperature-sensitive mutant fla10-1. This allele expresses a mutant kinesin-2 that allows one to rapidly switch-off anterograde IFT by shifting cells from the permissive (< 22°C) to the non-permissive temperature (32°C) for ~2 h [33]. In fla10-1 cilia, anterograde DRC4-GFP transport by IFT was observed at the permissive temperature of 16 °C but could not be observed after an incubation at 32 °C for >2.5 hours (Fig. 2e). Because DRC4-GFP transport is infrequent in steady-state cilia (Table S1), we used cells that were deciliated by pH shock and then allowed to assemble half-length cilia in the presence of cycloheximide (CHX) at 16 °C to determine whether the entry of DRC4-GFP is IFT-dependent. As described below in detail, high frequencies of anterograde DRC4-GFP transport (5.5 – 11.4 particles/min; Table S1) were observed in both wild-type and fla10-1 cilia of such cells at 16 °C (Fig. 2e). After shifting the CHX-treated cells to 32 °C for >2.5 hours, IFT transport of DRC4-GFP remained high in control cilia but was abolished in fla10-1 cilia (Fig. 2e and f; Table S1). Occasionally we observed DRC4-GFP particles that were undergoing diffusion in fla10-1 cilia (~0.3 particles/min; Fig. 2f) but the frequency of particles undergoing diffusion was ~18 × lower than the frequency of particles that were entering flagella by IFT in control cilia at the same temperature (Figs. 2e, f). These data indicate that the entry of DRC4-GFP into cilia is inefficient in the absence of IFT.

Fig. 2. DRC4-GFP entry into cilia is IFT dependent.

a, b) Phase contrast (a) and TIRF image (b) of a zygote with four cilia. Arrowheads, position of the basal bodies for one cilia pair. c) Kymogram of the cell depicted in a/b; the position of the basal bodies is indicated. d) Kymogram of a similar cell after photobleaching. Arrowheads, DRC4-GFP particles moving from the area of the basal bodies into the cilia. While often initially slower, the particles show linear trajectories indicative for active transport by IFT. Bars = 2 µm (a), 1 µm 1s (c, d).

e) Frequency of IFT transport of DRC4-GFP in steady-state cilia of fla10-1 and control cells and short cilia of CHX-treated cells at the permissive (16 °C) and non-permissive (32 °C) temperature. See Table S1 for additional controls.

f) DRC4-GFP transport in regenerated short cilia of CHX-treated fla10-1 at 16 ° and 32 °C. Bars = 2 µm and 1 s. T, position of the ciliary tip.

g) Mean square displacement vs. time for 71 DRC4-GFP particles diffusing inside steady-state cilia. The standard error of the mean at each value and a linear fit to the data (blue line) are indicated.

To address the question whether DRC4-GFP could distribute within cilia by diffusion, we determined the diffusion coefficient, D, of DRC4-GFP in steady-state cilia to be 0.110 µm²/s (± 0.019 µm²/s; Fig. 2g). A simple model for two-dimensional diffusion predicts that the concentration of DRC4-GFP will display a Gaussian distribution, $\rho =\frac{c}{2\sqrt{\pi Dt}}\text{exp}(-\frac{l^2}{4Dt})$, where l is the distance from the base of the cilium and C is the total number of particles. Thus, DRC4-GFP will be roughly evenly distributed along the length of the cilia when $\sqrt{2Dt}=12\mu \mathrm{m}$, the length of C. reinhardtii cilia. DRC4-GFP particles entering at the base of the cilium would therefore need on the order of 10 minutes to evenly distribute along cilia by diffusion [34]. This suggests that diffusion is too slow to move enough DRC4-GFP to the ciliary tip, in particular during ciliary growth when large amounts of DRC4 are required in a short time. We conclude that DRC4 requires active IFT for efficient transport across the transition zone barrier and rapid distribution inside cilia.

Spatial analysis of DRC4-GFP unloading from IFT

In many forms of intracellular transport, the loading and unloading of cargo are spatially regulated, typically by kinases, small GTPases, or calcium which can modulate the affinity of the carrier for the cargo [35]. It is currently assumed that cargoes are released from IFT particles upon their arrival at the tip. In steady-state cilia, the majority of DRC4-GFP particles (56%, n=93) were transported by anterograde IFT directly to the tip (Fig. 3j). At the tip, most of the DRC4-GFP particles remained stationary for an average of 1.7s (STD 0.8s, n=36) before moving away from the tip mostly by diffusion or, more rarely, by retrograde IFT (Figs. 3a, b, i, movie S5). Diffusion of IFT particle subunits has not previously been observed [21], and here we found that the diffusing DRC4-GFP particles were not associated with IFT20-mCherry (Fig. 3k). These observations imply that DRC4-GFP diffusion begins after dissociation from the IFT particle. At the tip, anterograde IFT particles need to be remodeled to become competent for retrograde IFT. Using IFT27-GFP, a dwell time of 2.5 s was observed for IFT particles to transition from anterograde to retrograde IFT; ~3.5 s were observed in Trypanosoma brucei using GFP-IFT52 as an IFT marker [36, 37]. Using IFT20-mCherry, we observed on average ~1.8 s (STD 1.0 s, n=14) required for IFT particle turn around at the ciliary tip. This time period is very similar to the dwell time of DRC4-GFP at the tip before the onset of diffusion, which suggests that the unloading of DRC4-GFP at the tip occurs late during the remodeling of IFT particles. The remodeling of the IFT particle probably alters its cargo binding properties causing a release of DRC4-GFP. We propose that remodeling of IFT particles at the tip represents a mechanism to unload cargoes.

Fig. 3. Spatial distribution of DRC4-GFP unloading sites.

a, b) Kymograms showing DRC4-GFP particles arriving at the ciliary tip (open arrowheads), and remaining stationary for ~2 s before the onset of DRC4-GFP diffusion (filled arrowheads). c, d) Kymograms showing DRC4-GFP particles directly transitioning from IFT to diffusion along the cilium. e–h) Kymograms showing diffusing DRC4-GFP putatively exiting cilia (a), recommencing anterograde (f) or retrograde (g) IFT, or stably docking to the axoneme (h). Arrowheads with T mark the position of the ciliary tip. Bars = 2 µm/1 s.

i) Histogram of the time periods in which DRC-GFP particles remained stationary after arriving at the ciliary tip by anterograde IFT.

j) Distribution of the sites at which DRC4-GFP terminated anterograde IFT based on 84 particles.

k) Kymograms showing that diffusing DRC4-GFP (green) is not associated with IFT20-mCherry (red). Bars = 1 µm/1 s.

A considerable number (~40%) of DRC4-GFP particles displayed IFT tracks terminating at various positions along the cilium (Fig. 3c, d, j, movie S6). Two-color imaging with IFT20-mCherry revealed that the IFT particles continued to move toward the tip after dissociation of DRC4-GFP (Figs. 1k). Thus, cargo unloading from IFT particles occurs also along the length of the cilium. After unloading from IFT trains at the tip or along the length of cilia, DRC4-GFP diffused for variable time and some particles apparently left cilia by diffusion (Fig. 3e). Some diffusing DRC4-GFP reinitiated anterograde or retrograde IFT at various sites along the cilium indicative of reloading onto IFT trains (Fig. 3f, g). This pattern of DRC4-GFP loading and unloading suggests that the spatial control of cargo capture and release by IFT particles is relaxed and probably regulated by a dynamic equilibrium between IFT particles, cargoes, and IFT-cargo complexes.

Real time analysis of DRC4-GFP assembly

After unloading from IFT, some DRC4-GFP particles became stationary, suggestive of the stable incorporation into the axoneme (Fig. 3h). We hypothesized that unloaded cargoes diffuse inside cilia until they are assembled onto their axonemal docking sites. If true, we predict that the average time of DRC4-GFP diffusion events should be reduced when empty DRC4 docking sites are abundant, e.g., in mutant cilia lacking DRC4. We mated the pf2 PF2-GFP cells to pf2-1 mutants (Fig. 4a). The resulting zygotes will have four cilia, two provided by the pf2-1 mutant parent initially devoid of DRC4 and DRC4-GFP (acceptor) and two derived from the PF2-GFP parent (donor) with occupied DRC4 docking sites (Fig. 4b–d). After photobleaching of the cilia, the impact of docking site availability on DRC4-GFP behavior inside cilia was analyzed. In acceptor cilia initially deficient of DRC4 and DRC4-GFP, DRC4-GFP diffused briefly after unloading from IFT and quickly docked to the axonemes (Figs. 4e: cilia 1 and 2, movie S7). In contrast, diffusion of DRC4-GFP lasted longer (10 s in PF2-GFP cilia, STD 6.3 s, n= 86 vs. 6.8 s, STD 5 s, n=51in pf2 cilia; t-test 0.001248) and docking was rare in photobleached donor cilia with occupied DRC4 docking sites (Fig. 4e: cilia 3 and 4, f, g). We conclude that DRC4-GFP diffuses to its axonemal binding sites after unloading from IFT particles and that the availability of free binding sites determines the duration of diffusion.

Fig. 4. DRC4-GFP diffuses to its axonemal assembly sites.

a) Schematic presentation of the C. reinhardtii mating reaction between a pf2 PF2-GFP donor and a pf2 mutant acceptor cell. b–d) Bright-field (b) and TIRF (c, d) images of a PF2-GFP × pf2 zygote before (c) and after photobleaching (d). Note the incorporation of DRC4-GFP predominantly near the tips of the mutant-derived acceptor cilia (nos. 1 and 2). Bar = 2 µm. e) Kymograms visualizing the movement of DRC4-GFP in the four cilia of the zygote depicted in b–d. Open arrows, diffusion; arrowheads, docking events. Note that diffusion of DRC4-GFP is reduced in mutant pf2 cilia lacking DRC4 (1 and 2) in comparison to cilia in which DRC4 docking sites are occupied with photobleached DRC4-GFP (cilia 3 and 4). Kymograms were selected from four 30-s recordings of the cell. Bars = 2 µm/1 s.

f) Comparison of the frequency of transport by IFT and diffusion of DRC4-GFP in pf2 acceptor and PF2-GFP donor cilia of chimeric zygotes. T-test values are indicated. g) Distribution of the time DRC4-GFP diffused in pf2 acceptor (solid bars; n=51) and PF2-GFP donor cilia (open bars; n= 86). Due to the abundance of diffusing DRC4-GFP in PF2-GFP donor cilia and the presence of residual unbleached DRC4-GFP reliable tracking of individual particles for extended time periods was difficult. Thus, many particles might actually diffuse longer in PF2-GFP donor cilia than indicated by the data, which are based on individual particles that were easy to track.

Replacement of axonemal DRC4 with newly imported DRC4-GFP is slow

In C. reinhardtii cilia, a subset of proteins is quickly replaced with newly synthesized proteins but both the sites of protein exchange in cilia and how exchange rate relates to protein supply are largely unknown [38]. To analyze the long-term incorporation rate of DRC4-GFP into cilia, zygotes were obtained by mating a DRC4-GFP donor strain to either pf2 mutants or wild-type cells (Fig. 5). DRC4-GFP rapidly accumulated in the initially DRC4-deficient acceptor cilia of PF2-GFP pf2 × pf2 zygotes [22]; the horizontal lines in kymograms obtained from such cilia (Fig. 5b) verified that DRC4-GFP is largely stationary indicative for stable incorporation (Fig. 5a–c). The incorporation was most prominent near in the distal region of the cilia but DRC4-GFP was also present scattered along the length of cilia. DRC4-GFP particles unloaded from IFT along the length of cilia attached to the axoneme without prior transport to the tip (Fig. 5c, movie S7). Thus, the pattern of DRC4-GFP incorporation during the repair of DRC4-deficient cilia reflects the distribution of its unloading sites from IFT.

Fig. 5. DRC4-GFP incorporation during ciliary repair and turnover.

Still images (a, d) and kymograms (b, c, e, f) of zygotes obtained by mating the PF2-GFP pf2 donor strain with either the pf2 mutant (a–c) or WT (d–f) acceptor cells. Note the high levels of DRC4-GFP incorporation into pf2 cilia (arrowheads in a) in comparison to WT cilia (filled arrows in d). *) corresponding kymograms are shown in b and e. DRC4-GFP is mostly stationary in pf2 cilia (b) while it is mobile in WT cilia with occupied docking sites. Arrowheads in e: retrograde IFT. The time since mixing of the gametes is indicated. The formation of zygotes typically peaks ~15 minutes after mixing of the gametes; older zygotes are likely to be enriched in samples analyzed >1 h after mixing of the gametes. c, f) Kymograms of bleached mutant-derived (c) or WT-derived (f) acceptor cilia. c) Kymogram showing anterograde IFT of two DRC4-GFP particles; unloading (open arrowheads) is followed by brief periods of diffusion before the particles stably incorporate into the mutant-derived acceptor cilium (solid arrowheads). The image corresponds to movie S7. f) DRC4-GFP incorporated into WT cilia bleaching in a single step (arrowheads) indicative for a single copy of the protein. Bars = 2 µm (a, d), 1 µm 1 s (b, c, e, f).

To analyze the exchange of DRC4 already anchored to the axoneme with newly imported DRC4, PF2-GFP pf2 donor cells were mated to wild-type acceptor cells in which DRC4 docking sites are occupied by the endogenous protein (movies S8 and S9). After 60–80 minutes, DRC4-GFP was incorporated near the ciliary tip of most WT-derived cilia (Fig. 5d, e). The distal end of the axoneme apparently undergoes cycles of shrinkage and growth which will generate free docking sites for DRC4-GFP near the tip [10]. Incorporation of DRC4-GFP along the length of wild-type cilia was rare and those signals mostly bleached in one step indicative for the presence of single molecules of DRC4-GFP (Fig. 5f). Most DRC4-GFP was diffusing inside WT-derived acceptor cilia (Fig. 5e) and the frequency of DRC4-GFP transport by retrograde IFT was increased to 0.5 particles/min (STD 0.16 particles/min, n= 3 experiments with 27 or more measurements each) compared to 0.1 particles/min observed in steady-state cilia of vegetative cells. Probably, the increased concentration of freely diffusing DRC4-GFP in such cilia promotes DRC4-GFP loading onto retrograde IFT particles.

A comparison of the total GFP signal present in cilia revealed that pf2 acceptor cilia (Fig. 5 a and b) retained considerably more DRC4-GFP than wild-type acceptor cilia (Fig. 5 d and e). The frequency of DRC4-GFP transport into zygotic cilia, however, was similar in PF2-GFP × pf2 and PF2-GFP × WT zygotes (~5 particles/min; see Table S2). We conclude that some DRC4-GFP transported into wild-type cilia must have left the cilia, probably by diffusion or retrograde IFT. In summary, the replacement of DRC4 anchored to the axoneme with DRC4-GFP imported de novo into such zygotic cilia is very slow.

DRC4-GFP transport during ciliary assembly

The motile and sensory functions of cilia require a precise control of organelle size, but the mechanism of ciliary length regulation is not well understood. Tubulin is continuously turned-over in the distal segment of cilia suggesting that the axoneme undergoes cycles of shrinkage and growth [12]. Thus, ciliary length is not static but controlled dynamically, which requires a steady supply of precursors and removal of disassembled material. Previous studies have shown that the amount and frequency of IFT proteins/particles in cilia is largely independent of ciliary length [39–41] but it is not known whether the amount of cargo transported by IFT is also independent of ciliary length. In photobleached steady-state cilia, the average frequency of DRC4-GFP transported by anterograde IFT was 0.65 particles/minutes. To determine the cargo frequency during ciliary growth, cilia were removed from cells by a pH-shock and allowed to regrow. In photobleached, regenerating cilia, GFP fluorescence was quickly restored to the tips, and many cargo particles arriving via IFT at the ciliary tip became stationary, indicative of stable incorporation into the axoneme (Fig. 6a–d). The average frequency of the anterograde DRC4-GFP transport in regenerating cells was 8.4 particles/minute (STD 2.2 particles/minutes based on three regeneration experiments each with ≥ 35 measurements). The transport frequency decreased linearly with increasing ciliary length (Fig. 6e, f). In summary, the frequency of DRC4-GFP transport by anterograde IFT particles was strongly increased during ciliary growth.

Fig. 6. DRC4-GFP transport is elevated during ciliary regeneration.

a–c) Cell with partially regenerated cilia (arrowheads) in phase contrast (a) and two frames from a recording after photobleaching of the cilia; the time points are indicated. Note recovery of DRC4-GFP fluorescence at the ciliary tip (arrowheads). d, e) Kymogram of DRC4-GFP transport in a regenerating cilium. Bars = 1 sec 1 µm.

e) Dot plot of the frequency of DRC4-GFP transport by anterograde IFT vs. ciliary length based on 46 recordings (represented by circles) of regenerating cilia. A trend line is shown. f) The frequency of DRC4-GFP transport in regenerating cilia of different length based on three independent regeneration experiments. The arrow indicates the average transport frequency in full length, steady-state cilia.

g) Western blots (bottom panels) probed with the antibodies indicated comparing cell bodies isolated from pf2 PF2-GFP cells with steady-state cilia (pre), regenerating cilia (post; ~35 min after pH-shock), and cilia regenerated in the presence of cycloheximide (post + CHX). In the latter, DRC4-GFP and IFT172 were reduced. Antibodies to PLD, a protein that is largely confined to the cell body [20], and Coomassie Blue staining (top) were used as loading controls. Similar to DRC4-GFP, the amount of IFT172 was reduced by ~50%. However, in vivo imaging of IFT20-mCherry in cilia of CHX-treated cells revealed normal to slightly reduced IFT frequencies.

h) Frequency of DRC4-GFP transport by anterograde IFT in steady-state cilia (pre), regenerating (post, early) and largely regenerated (late; −CHX) or growth-arrested (late; +CHX) cilia. In the absence of CHX-treatment, regenerating cilia were analyzed during regeneration (30–50 min after cilia amputation), and after cilia reach their steady-state length (80–120 min). CHX-treated cells were analyzed during (30–120 min after amputation) and after the assembly of short cilia (>240 min post deciliation). Note that in the absence of CHX the frequency of DRC4-GFP transport by anterograde IFT decreases when cilia reach their steady-state length while in the presence of CHX the transport frequency of DRC4-GFP remains elevated for hours in the short growth-arrested cilia. The data are based on one experiment; similar experiments are summarized in Table 1.

DRC4-GFP pool size and transport frequency

The expression of genes encoding ciliary proteins is induced after deciliation [42]. The increase in DRC4-GFP transport during regeneration could result from an increase in the amount of DRC4-GFP available in the cell body. The cytoplasmic pool could be depleted as cilia elongate lowering the concentration of available precursors and slowing ciliary growth. As suggested previously, limited pools of cytoplasmic precursors could control the ciliary length [43, 44]. Western blotting revealed that the amount of DRC4-GFP was similar in cell bodies isolated from cells with either steady-state or regenerating cilia (Fig. 6g, left). We conclude that DRC4-GFP synthesized de novo during ciliary regeneration only replenishes the cellular pool of DRC4-GFP. To test how a reduction in pool size affects the frequency of DRC4-GFP transport, cells were deciliated by a pH shock and allowed to regrow cilia in both the absence and presence of the protein synthesis inhibitor CHX (34). CHX-treated cells completed cilia regeneration in ~80 min to a length of ~7 µm (Table S1) resulting in a significant (>50%) reduction of DRC4-GFP remaining in the cell body (Fig. 6g, right). Despite the diminished pool of cellular DRC4-GFP, CHX-treated cells showed a high frequency of DRC4-GFP transport (~10 particles/min), similar to that of untreated cells during regeneration (Fig. 6h). CHX-treated cells maintained a high transport frequency of DRC4-GFP for hours while DRC4-GFP transport returned to steady-state levels in untreated control cells after cilia had reached their full-length of ~12 µm (Table S1, Fig. 6f). In the short growth-arrested cilia of CHX-treated cells, diffusing DRC4-GFP was abundant and the frequency of DRC4-GFP transport by retrograde IFT was increased (1.8 particles/min, STD 2.3 particles/min, n= 20 cilia), suggesting that some of the DRC4-GFP is exported (Fig. 2f). We conclude that the presence of short cilia triggers an increase of DRC4-GFP transport by IFT even when the amount of DRC4-GFP present in the cell body is reduced.

Discussion

The axoneme is organized into consecutive 96-nm structural repeats over most of its length and each 96-nm repeat contains a single N-DRC link [45]. This allows us to estimate the amount of DRC4 present in a full-length cilium. Assuming that one copy of DRC4 is present in each N-DRC linkage, the assembly of 12 µm-long cilia will require transport of ~1,000 copies of DRC4 (9 doublets each with ~120 of the 96-nm repeats each containing 1 DRC4) corresponding to an average transport frequency of 12.5 DRC4-GFP/min over the 80-min assembly period. DRC4-GFP entered growing cilia with an average frequency of ~8.4 units/min. The frequency was even higher in shorter cilia, and further extrapolation suggests an initial frequency of ~15 DRC4-GFP/min; the observed maximum was 34 DRC4-GFP/min. Sequential photobleaching of DRC4-GFP particles indicated that some IFT particles carry two copies of DRC4-GFP (Fig. S3). Thus, the experimental data are close to our estimates indicating that IFT transports DRC4-GFP in the amounts close to those expected for ciliary assembly.

Next, we tested to what extent our quantitative data on the DRC4-GFP transport fit with the observed kinetics of ciliary growth in C. reinhardtii [46]. We developed a rate equation to describe the growth of cilia that is in agreement with our observations and the previously published ciliary growth kinetics [46, 47]. The rate equation is $\frac{dL}{dt}=xr-x\beta$ with $x=\frac{12\mu m}{1000\mathit{\text{cargos}}}$ reflecting the change in cilia length per added DRC4-GFP unit. Our calculation is based on the assumption that the transport of DRC4-GFP is representative for that of other axonemal proteins including tubulin. Then each DRC4-GFP delivered will elongate cilia by 12 nm (i.e. 1000 DRC4 per cilium of 12 µm length). $r=\frac{15\text{cargos}}{60\text{sec}}(1-L/L_0)$ is the observed transport frequency of the DRC4-GFP which decreased linearly with increasing ciliary length (see Fig. 6f). L₀ is the cilia length at which the transport frequency drops to 0 and is a free parameter in our model (i.e. a variable parameter which is adjusted so that the model will predict a final ciliary length of 12 µm). A transport frequency of 0.65 DRC4-GFP/min was observed in steady-state cilia suggesting that this rate is sufficient to balance ciliary disassembly. Accordingly, we used $\beta =\frac{0.65\text{cargos}}{60\text{sec}}$ as the disassembly rate. For L₀ = 12.54 µm the final length of the cilia (i.e. the length for which $\frac{dL}{dt}=0$) is 12 µm. The L₀ value must be greater than the final cilia length in order to balance the disassembly rate. This equation results in an exponential growth function for the length with a time constant of 69.7 minutes; the cilia will reach 95% of its final length in approximately 200 minutes. This equation closely recapitulates the known kinetics of cilia growth in C. reinhardtii (Fig. 7).

Fig. 7. Differential cargo loading of IFT particles could regulate ciliary length.

Green line: Ciliary growth kinetics based on the differential cargo-loading model of IFT particles. The model suggests that IFT particles carry a large cargo load during assembly but are underutilized in steady-state cilia and that cargo loading onto IFT particles is regulated by ciliary length. Blue line: Ciliary growth kinetics based on the balance-point model in which cargo delivery is a function of the transit time of IFT particles while cargo loading is constant. Black dashed graph: Ciliary growth kinetics of C. reinhardtii as determined by Rosenbaum et al., 1969 [46].

Because the velocity of IFT particles is essentially constant, the time required to transit through cilia will increase as cilia elongate. This raises the question of whether the higher frequency of DRC4-GFP transport observed in short cilia represents an actual increase in the amount of cargo transported per IFT particle or simply reflects a given number IFT particles transporting the same amount of DRC4-GFP but making more tours through shorter cilia. The latter option is the basis of the balance-point model of ciliary length control, a model which requires neither a length sensor nor the regulation of the cargo load per IFT particle to establish cilia with a specific length [40, 41]: As the ciliary length increases, IFT particles will require more and more time to deliver cargoes to the tip resulting in a decreased frequency of cargo transport. When cargo delivery equals the loss of subunits at the tip, cilia will maintain a dynamic steady-state length. The balance point model can be described by the rate equation $\frac{dL}{dt}=\frac{xN}{\frac{5}{6}L+2}-x\beta$. The denominator reflects the transit time of IFT particles based on a dwell time of 2 s at the tip and 2 µm s⁻¹ anterograde and 3 µm s⁻¹ retrograde velocities. x is again the change in cilia length per cargo and β is the disassembly rate. N is the fractional loading of IFT particles (i.e. the percentage of IFT particles transporting DRC4-GFP) and is a free parameter in the model. For our calculations we used an IFT frequency of 60 particle/min [39, 40, 48]. A fractional loading of N=0.13 cargos will result in a final length of 12 µm, but cilia would need many hours to reach this length (Fig. 7). Thus, changes in the transit time of IFT particles are insufficient to explain the difference in the frequencies of DRC4-GFP transport observed in steady-state vs. regenerating cilia. DRC4 is only one of several hundred proteins which are transported into the cilium. Extrapolation of our data on DRC4-GFP suggests that IFT particles in short regenerating cilia carry a significantly larger cargo load than those in steady-state cilia.

The expression of genes encoding ciliary proteins is induced during ciliary regeneration [42, 49] and a higher concentration of precursors in the cell body could increase the amount of ciliary proteins transported by individual IFT particles. However, when we partially depleted the cellular pool of DRC4-GFP by CHX-treatment the frequency of DRC4-GFP entry into the short growth-arrested cilia assembled by such CHX-treated cells was unaffected suggesting that a high cargo loading of IFT can be achieved even when the size of the cellular cargo pool was lowered. We propose a differential cargo loading model in which cells sense the presence of truncated cilia and respond by increasing the amount of cargo loaded onto IFT particles until the normal ciliary length is established. Alternatively, the admission of highly loaded IFT particles through the transition zone into the cilium could be regulated in a length-dependent manner.

Putative candidates for such a length-dependent sensor include the various protein kinases which function in ciliary length regulation [50]. The phorphorylation level at a site in the activation loop of the Chlamydomonas Aurora-like Kinase (CALK) is proportional to ciliary length during assembly [51]. It is currently unknown whether the ciliary length-dependent increases in CALK activity and cargo transport by IFT are functionally linked. Earlier observations indicated that the amount of cargo in cilia is reduced in situations in which cells need to shorten or resorb their cilia supporting our notion that the amount of cargo transported by IFT is regulated [52]. Further, anterograde IFT particles move somewhat slower during ciliary regeneration [40]; this could be caused by an increased drag due to the larger load of (transmembrane) proteins carried by IFT particles in short cilia. Note that our model can be reconciled with the balance point model. There is strong evidence that the amount of IFT particles and proteins is stable in cilia of different length [40, 41]. Thus, a limited supply of IFT materials could limit the growth rate, especially in short cilia that grow rapidly. However, as the cilia grow longer, the cargo capacity of IFT trains may be underutilized and a sensor-based cargo loading mechanism could be the primary determinant of the assembly rate. This solution could allow for rapid fine tuning of the length of cilia during the cell cycle and in response to environmental cues.

Highlights.

In vivo imaging reveals that the axonemal protein DRC4 is a cargo of IFT

DRC4-GFP dissociates from IFT particles at various sites along the cilium

Unloaded DRC4-GFP diffuses to its axonemal docking sites

IFT particles carry a larger cargo load during ciliary growth

List of abbreviations

CHX

cycloheximide

N-DRC

nexin-dynein regulatory complex

IFT

intraflagellar transport

PF/pf

paralyzed flagella

GFP

green fluorescent protein

PLD

phospholipase D

WT

wild type